Background
Prostate cancer is the most common cancer in men in Western societies [
1]. At diagnosis the majority of prostate cancer is curable, but a minor subset of tumors is characterized by aggressive growth and metastasis. Despite recent advance in research for biomarkers, the established pre-treatment prognostic parameters are Gleason score, tumor extent on biopsy, pre-operative PSA and clinical parameters. These data are statistically powerful but not sufficient for optimal individual treatment choice. It is therefore hoped that the analysis of further biomarkers may lead to improved individual prediction of tumor aggressiveness in the future.
Breast cancer anti-estrogen resistance 1 (BCAR1/p130Cas) is a scaffold protein that serves as a hub in cellular signaling. It facilitates the assembly of multi-protein complexes regulating diverse cellular processes such as migration, invasion, proliferation and survival. BCAR1 participates in signal conduction of major oncogenic kinases such as Abl, FAK and Src. Consequently, BCAR1 has been shown to be overexpressed in diverse malignancies, including cancers of the breast, lung, liver and brain, and has been linked to adverse features in these entities (reviewed in [
2,
3]). Initial evidence also suggests a role for BCAR1 in prostate cancer progression, as its overexpression was linked to an unfavorable tumor phenotype and biochemical relapse in three studies analyzing 110 [
4], 130 [
5] and 242 [
6] prostate cancer specimens.
Based on these data, we intended to confirm the biologic and prognostic role of BCAR1 protein in a very large cohort of prostate cancer patients. For this purpose, we chose our tissue microarray (TMA) comprising >11,000 prostate cancer specimens with attached clinical and molecular data. Our study highlight that BCAR1 expression is associated with unfavorable tumor features and that the prognostic impact of BCAR1 is limited to ERG-negative cancers.
Methods
Patients
Radical prostatectomy samples were taken from 11,152 patients, undergoing surgery between 1992 and 2011 at the Department of Urology and the Martini Clinics at the University Medical Center Hamburg-Eppendorf. Follow-up was available from 9695 patients (median 36.8 months; range 1 to 228 months; Additional file
1: Table S1). Prostate specific antigen (PSA) recurrence was defined as a postoperative PSA of ≥0,2 ng/ml. Histological analysis of prostate specimen was done as detailed in [
7] and TMA were produced as described earlier in [
8]. Each TMA block contained various control tissues, including normal prostate tissue. The molecular database attached to this TMA contained results on Ki67 expression in 7010 (expanded from [
9]), ERG expression in 9628, ERG break apart fluorescence in-situ hybridization (FISH) analysis in 6106 (expanded from [
10]), and deletion status of 5q21 in 3037 (expanded from [
11]), 6q15 in 3528 (expanded from [
12]), PTEN in 6130 (expanded from [
13]), and 3p13 in 1290 (expanded from [
14]) tumors.
Immunohistochemistry
Freshly cut TMA sections were stained on 1 day and in one experiment. Slides were deparaffinized and exposed to heat-induced antigen retrieval at 121 °C in Tris-EDTA-citrate buffer (pH 7.8). BCAR1 specific mouse monoclonal antibody (clone M144, Abcam, Cambridge, UK) was applied at 1/37.5 dilution at 37 °C for 60 min. BCAR1 staining was visualized with the EnVision Kit (Dako, Glostrup, Denmark) according to the manufacturer’s directions. Staining was localized to the cytoplasm. It was homogenous in the analyzed tissue samples and therefore staining intensity was semi quantitatively assessed as negative, weak, moderate, and strong.
Statistics
JPM 9 software (SAS Institute Inc., NC, USA) was used. Contingency tables and the likelihood-ratio chi2-test were performed to find associations between molecular parameters and clinical tumor characteristics. Kaplan-Meier survival curves were calculated and the Log-Rank test was applied to detect differences. Cox proportional hazards regression was performed to look for statistical independence of pathological, molecular and clinical variables. Separate analyses were done using various sets of parameters available either before or after prostatectomy.
Discussion
The results of this study suggest that high BCAR1 expression is a weak independent predictor of unfavorable tumor characteristics and early PSA recurrence. Consistent with earlier findings [
5], BCAR1 staining was barely detectable in luminal cells of non-neoplastic prostatic glands but clearly up regulated in a large fraction of prostate cancers, which suggests a role for BCAR1 during prostate cancer development. The analysis revealed cytoplasmic BCAR1 staining in 76.6% of 9495 analyzable prostate cancers. These numbers fit well to previous TMA based studies, which reported up to 90% of BCAR1 positivity in sets of 110 up to 242 prostate carcinomas [
4‐
6].
BCAR1 up regulation was linked to aggressive cancer features in our study, including high Gleason score, advanced tumor stage, presence of lymph node metastases, rapid tumor cell proliferation and early biochemical recurrence, arguing for a contribution of elevated BCAR1 protein expression to prostate cancer progression. These findings are supported by the results of several earlier studies suggesting associations between BCAR1 up regulation and advanced prostate cancer features such as castration resistance, metastasis and early biochemical relapse [
4‐
6]. A tumor promoting role of BCAR1 overexpression fits well to the known function of BCAR1, which serves as a hub for several oncogenic pathways regulating processes like cell proliferation, growth, migration, and other cancer relevant cellular functions (reviewed in [
2,
3]).
The molecular database attached to our TMA allowed us to draw conclusions on molecular mechanisms associated with BCAR1 up regulation. It is well known that about the half of prostate cancers carry a gene fusion, which links the androgen-regulated serine protease TMPRSS2 with the ETS-transcription factor ERG resulting in an androgen-related expression of ERG with subsequent dysregulation of more than 1600 ERG target genes [
17‐
19]. BCAR1 up regulation was strongly linked to TMPRSS2:ERG fusions in our study. ERG does not seem to be implicated in transcriptional control of BCAR1 based on the results of studies analyzing global transcriptional changes between ERG-negative and ERG-positive prostate cancers [
15,
19‐
22]. It is, thus, possible that post-transcriptional modifications may account for the different BCAR1 expression levels in ERG-positive and ERG-negative cancers, including for example altered protein stability. This assumption is supported by studies demonstrating that ERG activation modulates the activity of the reversible protein ubiquitination cascade [
15,
23], including the E3 ubiquitin ligase SKP2 that regulates stability of the BCAR1/p130CAS protein [
24‐
26].
The association between BCAR1 expression and lymph node metastasis was puzzling because it was in the opposite direction between the ERG negative and positive cancers and significant in both subsets (Additional file
1: Table S2). We assume a complex situation as a result of 1) minor differences resulting in high statistical significance because of the very high sample numbers, 2) ERG driven sudden up-regulation of BCAR1 expression and 3) regression to the mean in tumor progression to nodal metastasis which means increase of BCAR1 staining in the
ERG negative and decrease of BCAR1 staining in the
ERG positive cancer subset.
In prostate cancer are after the TMPRSS2: ERG fusion, chromosomal deletions the most frequent type of genomic aberration. They occur at frequencies of up to 40% [
15,
27] and are associated with poor prognosis [
11‐
15]. Deletion of PTEN (20%), 6q (20%), 5q (10%) and 3p (10%) are linked to either positive ERG status (PTEN, 3p) or negative ERG status (6q, 5q). In our study, BCAR1 up regulation was associated with most of these deletions (PTEN, 5q21, and 6q15). This finding is consistent with earlier work linking altered BCAR1 activity to development of genetic instability [
28]. It has been shown that BCAR1 can translocate to the nucleus under hypoxic conditions, where it specifically impairs the homologous repair (HR) protein RAD51 [
28]. Several studies have demonstrated that RAD51 deficiency can induce replication defects, genetic instability and chromosomal rearrangements [
29,
30].
In previous studies, we identified several proteins, which were also expressed at higher levels in ERG-positive than in ERG-negative prostate cancer. In some of these, the prognostic effect was likewise restricted to the ERG-negative subset [
31‐
33]. Here we identify BCAR1 as a protein following this pattern. In opposite some other biomarker were only prognostic in ERG-positive cancer [
32,
34]. Together, these data show that tumor relevant functions of BCAR1 and other proteins turn out to be attenuated or amplified by ERG. ERG seems to be a critical modifier of the intracellular environment [
19,
20,
23]. These challenge the concept of a unique prognostic molecular test applicable to all patients [
35,
36]. It appears possible that different tests need to be developed for ERG-positive and ERG-negative cancer. Furthermore, the small difference of about 10% in Kaplan-Meier plots between negative and strongly positive BCAR1 expression, shows that BCAR1 seems to be a weak prognostic marker (Fig.
4). Thus the BCAR1 biomarker may best aid in decision making if combined with other marker in ERG-negative prostate cancer.
It is of note that the Gleason score had the highest hazard ratio for PSA recurrence-free survival in multivariate analysis and is therefore the strongest (and less expensive) prognostic marker in prostate cancer. We demonstrated recently, that with the percentage of unfavorable Gleason patterns, Gleason grading could be transformed from a categorical into a continuous variable (i.e., the quantitative Gleason score) with subtler distinction of prognosis [
37]. The power of morphological methods competing with biomarkers for predicting prostate cancer aggressiveness is best demonstrated by the separate analysis of different prognostic Gleason groups. That the prognostic impact of BCAR1 was lost in groups defined by classical Gleason score categories or by the quantitative Gleason score demonstrates how difficult it is for a biomarker to outperform a morphological malignancy score.